OGB-LSC (OGB Large-Scale Challenge)

OGB 大规模挑战赛 (OGB-LSC) 是三个真实世界数据集的集合，用于推进大规模图 ML 的最新技术。 OGB-LSC 提供的图数据集比现有的大几个数量级，涵盖了三个核心的图学习任务——链接预测、图回归和节点分类。 OGB-LSC 由三个数据集组成：MAG240M-LSC、WikiKG90M-LSC 和 PCQM4M-LSC。每个数据集提供一个独立的任务。 MAG240M-LSC 是一个异构学术图，任务是预测位于异构图中的论文的主题领域（节点分类）。 WikiKG90M-LSC 是一个知识图谱，任务是估算缺失的三元组（链接预测）。 PCQM4M-LSC 是一个量子化学数据集，任务是预测给定分子的一个重要分子特性，即 HOMO-LUMO 间隙（图回归）。

The Open Graph Benchmark Large-Scale Challenge (OGB-LSC) is a collection of three real-world datasets dedicated to advancing the state-of-the-art in large-scale graph machine learning. The graph datasets provided by OGB-LSC are several orders of magnitude larger than existing counterparts, covering three core graph learning tasks: link prediction, graph regression, and node classification. OGB-LSC consists of three datasets: MAG240M-LSC, WikiKG90M-LSC, and PCQM4M-LSC, each corresponding to an independent task. MAG240M-LSC is a heterogeneous academic graph, with the task of predicting the subject areas of papers within this heterogeneous graph (node classification). WikiKG90M-LSC is a knowledge graph, tasked with inferring missing triples (link prediction). PCQM4M-LSC is a quantum chemistry dataset, whose task is to predict an important molecular property of given molecules, namely the HOMO-LUMO gap (graph regression).

搜集汇总

背景与挑战

背景概述

OGB-LSC是一个包含三个大规模图数据集（MAG240M-LSC、WikiKG90M-LSC和PCQM4M-LSC）的集合，用于推进图机器学习技术，涵盖节点分类、链接预测和图回归三大核心任务。该数据集由斯坦福大学等机构于2021年发布，旨在提供比现有数据集更大规模的图数据以支持研究。